Papers by Oleh Tereikovskyi
Management of Development of Complex Systems
The relevance of the implementation of means of recognition of the emotional state by the image o... more The relevance of the implementation of means of recognition of the emotional state by the image of the face into the personnel management system is well-founded. It is shown that the implementation of such tools leads to the need to adapt the values of architectural parameters of neural network models for detecting the boundaries of target objects on bitmap images to the expected conditions of use. An approach to determining the most effective type of neural network model is proposed, which involves expert evaluation of the effectiveness of acceptable types of models and conducting computer experiments to make a final decision. As a result of the conducted research, it was determined that among the types of neural network models tested in the task of segmentation of raster images, the U-Net model is the most effective for detecting facial borders on small raster images. Using this neural network model provides a mask selection accuracy of 0.88. At the same time, the necessity of imp...
The Method of Semantic Image Segmentation Using Neural Networks
International Journal of Image, Graphics and Signal Processing
Procedure for Using Neural Networks for Segmentation of Raster Images
Cybersecurity: Education, Science, Technique
Currently, means of semantic segmentation of images, based on the use of neural networks, are inc... more Currently, means of semantic segmentation of images, based on the use of neural networks, are increasingly used in computer systems for various purposes. Despite significant successes in this field, one of the most important unsolved problems is the task of determining the type and parameters of convolutional neural networks, which are the basis of the encoder and decoder. As a result of the research, an appropriate procedure was developed that allows the neural network encoder and decoder to be adapted to the following conditions of the segmentation problem: image size, number of color channels, permissible minimum accuracy of segmentation, permissible maximum computational complexity of segmentation, the need to label segments, the need to select several segments, the need to select deformed, displaced and rotated objects, the maximum computational complexity of learning a neural network model is permissible; admissible training period of the neural network model. The implementati...
Speaker's Emotions Recognition Module Based on the GoogleLeNet Neural Network
2022 International Conference on Smart Information Systems and Technologies (SIST)
Scientific notes of Taurida National V.I. Vernadsky University. Series: Technical Sciences
Determination of Signs of Information and Psychological Influence in the Tone of Sound Sequences
2020 IEEE 2nd International Conference on Advanced Trends in Information Theory (ATIT), 2020
The article is devoted to the development of information technology for recognizing the destructi... more The article is devoted to the development of information technology for recognizing the destructive influences in the non-verbal content of the sound sequences of multimedia messages of Internet-oriented mass media. The necessity of determining the significant signs of informational and psychological influence in the tonality of the sound series is introduced. It was found that, first of all, such characteristics of music as pulsation, a binaural effect and changes in the main tone should be analyzed. It is proposed to determine the signs of information-psychological influence by analyzing the music used in multimedia messages, which destructive properties have been proven. The object of the study is the music of the song ≪Не хочу ни любви, ни почестей≫ (‘I want neither love nor honor’), which is used as a video background in the famous suicidal game ≪Синий кит≫ (‘Blue Whale’). To analyze the tonality of music on each quasi-stationary fragment, the time variation of the modulus, the real part, and also the imaginary part of the complex representation of the Fourier coefficient was calculated. During the calculations, the length of the quasi-stationary fragment was 1024 samples, which at a sampling rate of 44100 Hz corresponds to about 20 ms. As a result of the analysis, it was determined that significant signs of information and psychological influence are the low-frequency nature of the sound, the periodic nature of the amplitude-time indicators of the sound signal with a period of 5–12 seconds, the periodic nature of the amplitude-frequency indicators of the fundamental tone of the sound signal with a period of 5–7 seconds. To calculate and visualize these indicators, a software package specially developed in the MatLab 2018 environment was used. It is proposed to correlate the ways of further research with the development of a method for recognizing the presence of destructive influences in the non-verbal content of the sound sequences of multimedia messages distributed in Internet-oriented mass media.
Keyboard Dynamic Analysis by Alexnet Type Neural Network
2020 IEEE 15th International Conference on Advanced Trends in Radioelectronics, Telecommunications and Computer Engineering (TCSET), 2020
In this article has been reviewed questions of development neural network analysis tools of keybo... more In this article has been reviewed questions of development neural network analysis tools of keyboard handwriting indicators for personality and user emotions recognition. Installed ability to upgrade specified funds through the use of convolutional neural networks of AlexNet type, which makes it necessary to evaluate the effectiveness of such use. It was also determined that it is possible to evaluate the efficiency of using the neural network model experimentally with using indicators of recognition accuracy and duration of training. A software implementation of AlexNet was developed, and a training sample was formed, consisting of 1005 examples of the parameters of the dynamics of keyboard handwriting for 10 users. As parameters characterizing of the dynamics of keyboard handwriting has been used holding time and the time between successive pressing of two keys. Using computer experiments, it was found that in a fairly limited training sample at 50 training epochs, AlexNet allows achieving user recognition accuracy of over 80%, which is comparable to the results of the best modern systems of similar purpose and confirms the possibility of effective use of this type of network for analyzing the dynamics of keyboard handwriting. The need for further research in the direction of the formation of the training sample, that providing high-quality training of the neural network model is shown. The expediency of developing a method for adapting AlexNet architectural parameters to the conditions of the task of analyzing the dynamics of keyboard handwriting was also determined.
Parameter Definition for Multilayer Perceptron Intended for Speaker Identification
2019 IEEE International Conference on Advanced Trends in Information Theory (ATIT), 2019
The article deals with the improvement of speaker identification tools. The prospects of neural n... more The article deals with the improvement of speaker identification tools. The prospects of neural network identification tools are established. Authors show that ways of improving such identification tools are associated with the adaptation of the neural network model used to the significant conditions of the task. The authors propose to increase the efficiency of neural network identification tools by adapting the parameters of a deep neural network with direct signal propagation to such conditions of the speaker identification task as the parameters of the voice signal, the number of identified speakers and training samples. The adaptation approach providing for the experimental definition of multilayer perceptron structural parameters is developed. The identification issue under study involved: the number of speakers – 10, voice signal duration – 7. 008s, voice signal sampling frequency 16,000 Hz, quasi-stationary fragment duration–0.016 s, and the number of Mel-frequency cepstral coefficients (MFCC) of the single quasi-stationary fragment – 20. The training sample includes 100 recordings of the voice signal for 10 speakers when they read texts in English. Recording took place in studio. Each fragment of the voice signal sample is unique. The neural network model adapted to these conditions is a three-layer perceptron with 256 neurons in the first hidden layer and 80 neurons in the second hidden layer. The ReLU activation function is used for the hidden layer neurons, and the Softmax activation function is used for the output neurons. Given an acceptable level of resource intensity, the developed neural network model allows achieving identification accuracy of about 0.95, which is comparable with the most modern means of a similar purpose. The necessity of further research on developing a method of adaptation of multilayer perceptron architectural parameters to the task of integral identification of speaker emotions and personality is substantiated.
International Journal of Modern Education and Computer Science, 2021
The problem of the article is related to the improvement of means of covert monitoring of the fac... more The problem of the article is related to the improvement of means of covert monitoring of the face and emotions of operators of information and control systems on the basis of biometric parameters that correlate with twodimensional monochrome and color images. The difficulty in developing such tools has been shown to be largely due to the cleaning of images associated with biometric parameters from typical non-stationary interference caused by uneven lighting and foreign objects that interfere with video recording. The possibility of overcoming these difficulties by using wavelet transform technology, which is used to filter images by combining several identical, but differently noisy monochrome and color images, is substantiated. It is determined that the development of technology for the use of wavelet transforms is primarily associated with the choice of the type of basic wavelet, the parameters of which must be adapted to the conditions of use in a particular system of covert monitoring of personality and emotions. An approach to choosing the type of basic wavelet that is most effective in filtering images from non-stationary interference is proposed. The approach is based on a number of the proposed provisions and efficiency criteria that allow to ensure when choosing the type of basic wavelet taking into account the significant requirements of the task. A filtering procedure has been developed, which, due to the application of the specified video image filtering technology and the proposed approach to the choice of the basic wavelet type, allows to effectively clean the images associated with biometric parameters from typical non-stationary interference. The conducted experimental studies have shown the feasibility of using the developed procedure for filtering images of the face and iris of operators of information and control systems.
Uploads
Papers by Oleh Tereikovskyi